summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2011-06-29asdfHEADnewuiSøren Sandmann Pedersen1-2/+4
2011-06-29Improved scrollingSøren Sandmann Pedersen2-44/+56
2011-06-29Get rid of screenshot windowSøren Sandmann Pedersen3-383/+6
2011-06-29TodoSøren Sandmann Pedersen1-2/+20
2011-06-29Delete some unused codeSøren Sandmann Pedersen2-103/+3
2011-06-29todoSøren Sandmann Pedersen1-0/+10
2011-06-29More GtkTreeView performance hacks.Søren Sandmann Pedersen2-110/+98
On x86-64 GLib is buggy in that g_atomic_get_pointer() ends up taking a mutex instead of just using atomic instructions. This makes GValue handling really slow, so avoid that by directly comparing the contents.
2011-06-29Apply yet another workaround for GTK+ performance issues.Søren Sandmann Pedersen1-4/+24
Sorting ends up emitting rows_reordered a lot, which requires building a bunch of GtkTreePaths, which is O(n^2) in the number of indices. By temporarily unsetting the model, we can avoid emitting those signals. To avoid an ugly flash, we also temporarily disable updating the tree view window. See https://bugzilla.gnome.org/show_bug.cgi?id=634491
2011-06-29Eliminate caller viewSøren Sandmann Pedersen3-213/+78
2011-06-29Add callers linesSøren Sandmann Pedersen1-23/+29
2011-06-29Descendants lineSøren Sandmann Pedersen2-18/+35
2011-06-29Expand on double-clickSøren Sandmann Pedersen2-19/+56
2011-06-29Make sure there is always a caller in the callers view.Søren Sandmann Pedersen1-1/+21
This hacks around a GTK+ bug where the initial size allocation is screwed up for empty tree views.
2011-06-29Hook up selectionSøren Sandmann Pedersen1-8/+6
2011-06-29Reorder columnsSøren Sandmann Pedersen3-8/+15
2011-06-29Add everything backSøren Sandmann Pedersen1-0/+4
2011-06-29Fix bug in move to frontSøren Sandmann Pedersen1-8/+12
2011-06-29Initial WIP new uiSøren Sandmann Pedersen3-157/+39
2011-06-29Put everything in one big treeSøren Sandmann Pedersen2-10/+20
2011-01-27cli: Allow to get samples from single pidDamien Lespiau4-8/+30
perf lets you decide to only get events that concerns a single process and thus make sysprof profile this process instead of the whole system (it can happen that you don't really care about other processes that are just then noise). As a side effect, this allows sysprof to not run as root if you have the rights on the process you want to profile.
2011-01-27tracker: Silence a debug messageDamien Lespiau1-0/+2
The same way than the other ones, with #if 0.
2011-01-27Add a .gitignore fileDamien Lespiau1-0/+29
Hooray for a clean git status!
2011-01-27Speed up map locating in tracker.cSøren Sandmann Pedersen1-53/+47
Store the maps as a GArray rather than a GPtrArray to improve locality; also move maps to front when they are looked up.
2011-01-27Don't compute CRC32s that we know will fail.Søren Sandmann Pedersen1-3/+19
If two files both have build-ids and those build-ids don't match, there is no point computing the CRCs.
2011-01-27footreestore: Eliminate some type checks that were showing up on profilesSøren Sandmann Pedersen1-0/+8
2011-01-27Call gdk_flush() in set_busy().Søren Sandmann Pedersen1-4/+2
Otherwise, the cursor won't show up. (GTK+ 2.22).
2011-01-27Right align the samples labelSøren Sandmann Pedersen1-1/+1
2011-01-25build: Update autotools configuration a bitJavier Jardón3-7/+16
2011-01-24Remove INSTALL file, It's autogeneratedJavier Jardón1-182/+0
2011-01-24autogen.sh: Use autoreconfJavier Jardón1-76/+11
2010-05-06Post-release version bump to 1.1.7Søren Sandmann Pedersen1-1/+1
2010-05-06Pre-release bump to 1.1.6Søren Sandmann Pedersen1-1/+1
2010-05-06Makefile.am: barrier.h is now called util.hSøren Sandmann Pedersen1-1/+1
2010-04-24Fix use of uninitialized variableSøren Sandmann Pedersen1-2/+1
2010-04-24Use the SET_OUTPUT ioctl to direct all output to the same bufferSøren Sandmann Pedersen2-30/+53
This ensures that we don't get events out of order, which will make the profiles look a lot less bogus on multi-core systems.
2010-04-24Make comm events reuse any existing process.Søren Sandmann Pedersen1-43/+52
A comm event doesn't necessarily mean a new process was created. It may just mean that an existing process changed its command line.
2010-04-23Eliminate some compiler warningsSøren Sandmann Pedersen5-12/+20
2010-04-23Use d_print() instead of g_print() for debug spamSøren Sandmann Pedersen1-1/+1
2010-04-23Add FIXME about performance issue in stack_stash_add_trace()Søren Sandmann Pedersen1-0/+7
2010-04-23sysprof-cli: Print out number of samples when savingSøren Sandmann Pedersen1-1/+3
2010-03-23Fix encoding of error messagePascal Terjan1-1/+1
Use g_strerror instead of strerror to get UTF-8
2010-03-23Fix build with -Werror=format-securityPascal Terjan1-1/+1
Don't use dynamic string as format, one of the args could be a string containing %.
2010-03-21Really disable hardware counters.Søren Sandmann Pedersen1-3/+4
2010-03-10Ignore fork events that are just new threads being spawned.Søren Sandmann Pedersen2-2/+13
These would cause mysterious new processes to show up.
2010-03-09Disable use of hardware perf counters.Søren Sandmann4-6/+10
They don't seem to actually work on a lot of hardware.
2010-03-09Stop using double mmap trick.Søren Sandmann1-34/+46
It has issues on various architectures, such as ARM, and there was never any guarantee that it would actually work. So since we'd need fallback code anyway, and the main point of it was simplifying the code, just get rid of it and handle the overflow conditions manually.
2010-02-20Update list of kernel symbols to take counter->event rename into account.Søren Sandmann Pedersen1-0/+6
2010-02-08Coding style fixSøren Sandmann Pedersen1-1/+2
2010-02-08Wire up __NR_perf_counter_open syscall for almost all arches as ofNix1-1/+35
Currently, sysprof hardwires the syscall number of __NR_perf_counter_open. That's all very well, except... like other syscall numbers, it varies per-arch. With a few exceptions (e.g. sparc64), it even varies between 32-bit and 64-bit x86 arches. This patch provides the defines for every number I could see assigned in git trunk right now, with the exception of some of the ARM ones, where I couldn't see how to differentiate between the various sorts of ARM to figure out what the syscall base was. For those, let's wait for libc to catch up. The only arch I've been unable even to guess at is the very new S-Core, which is the only user so far of the generic syscall table in <include/asm-generic>. This specifies a value of 241, but I can't hook it up because I can't find an S-Core toolchain anywhere so I don't know what #define to check for, and googling gives no clues. Only tested on x86-64 (where it now works) so there may be all sorts of bugs in the rest of it (well, in the #defines anyway, though most were checked against GCC trunk).
2010-02-01Unconditionally fall back to software if opening hardware counters failsswSøren Sandmann Pedersen1-8/+5
If hardware counters are not available, fall back to SW regardless of why the hardware counters aren't available. The error code can be either ENOTSUPP or ENODEV depending on CPU type, so it's simpler to just retry in all cases, and only fail if the software fallback failed.
generated by cgit v1.2.3 (git 2.39.1) at 2024-05-14 03:38:38 +0000